I was interested to read a randomised controlled trial of pulsed xenon UV (PX-UV) for room disinfection in Clinical Infectious Diseases. In this study, PX-UV was not associated with a reduction in HCAI. Some may be quick to say that this provides evidence that “automated room disinfection systems don’t work”, but there are important differences in the technologies involved as well as the usual strengths and limitations of the study methodology that could explain this outcome.

There are some convincing studies that automated room disinfection systems – both those based on hydrogen peroxide and on UVC – are effective in reducing HCAI. For example, I was involved in a study in the US showing that patients admitted to rooms disinfected using hydrogen peroxide vapour were less likely to acquire multidrug-resistant organisms (MDRO) than those admitted to rooms disinfected using standard methods. And then there’s the BETR-D studies, showing that UVC is associated with reduced MDRO acquisition and HCAI. So, there was a compelling rationale for expecting this study of PX-UV to show similar reductions in HCAI.

The study itself was a sham controlled cluster randomised cross-over trial of PX-UV with 15 wards across two US hospitals randomised to either PX-UV treatment or sham control for terminal disinfection of all patient rooms. The primary outcome was the identification of organisms associated with both HCAI and the environment from patients >3 days after hospital admission. More than 25,000 patients were included, and there were no significant differences in the incidence of HCAI in the intervention and control arms (3.49 vs 3.17 infections/1000 patient-days, respectively).

Some points for discussion:

• Why didn’t we see a reduction in HCAI? Could have been a number of different reasons for this. If we were to rank automated room disinfection systems by ‘microbiological effectiveness’, you would say hydrogen peroxide vapour > hydrogen peroxide aerosol > UVC > PX-UV. So it could be that the microbiological effectiveness of the system was not sufficient to reduce HCAI risk. Also, the definition of “HCAI” was fairly unique in this study, with certain organisms linked with the environment selected, and a rather long 3 day exclusion at the start of the hospital stay where the identification of these organisms was not considered “healthcare-associated”. So the definition of HCAI could have been a factor in the observed outcomes. Related to this, there was no measurement of MDRO acquisition based on active surveillance cultures. This is an important factor, because many HCAI are endogenous and unlikely to be affected by a terminal disinfection intervention (as illustrated nicely by the accompanying editorial).
• No microbiological sampling of the environment was undertaken, making it difficult to qualify actual differences the microbial risk profile from surfaces.
• The addition of a sham control is a nice touch. But I am not sure how valuable it is in the context of this study. Is knowing how a room is disinfected really likely to impact behaviour in a way that will influence HCAI risk? Perhaps if ward teams became invested in an intervention (or actively wanted it to fail!) then having the sham control would reduce the risk of that bias influencing the findings?

This helpful and well-designed study certainly raises some important questions about automated room disinfection technologies. It calls for further comparative effectiveness evaluations so that we can compare different technologies with the same methodology. The reality is that comparative effectiveness studies with clinical outcomes – say, for example, a cluster randomised trial with UVC in one arm and PX-UV in another – would be a very difficult undertaking. So, I think we’ll be left with the usual patchwork of study designs when choosing our optimal approach to reduce HCAI and maximise patient safety.